Nonlocal structural perturbations in a mutant human insulin

The Mode of Interaction of the Relaxin-like Factor (RLF) with the Leucine-rich Repeat G Protein-activated Receptor 8. Erika E. B?llesbach , Christian ...
0 downloads 0 Views 1MB Size
Download PDF
Biochemistry 1992,31, 11940-1 1951

11940

Nonlocal Structural Perturbations in a Mutant Human Insulin: Sequential Resonance Assignment and 13C-Isotope-Aided 2D-NMR Studies of [PheB24+Gly]Insulin with Implications for Receptor Recognition7 Qing Xin Hua,bs Steven E. Shoelson,ll and Michael A. Weiss’JJ Department of Biological Chemistry and Molecular Pharmacology, Haruard Medical School, Boston, Massachusetts 02115, J o s h Diabetes Center and Department of Medicine, Brigham and Women’s Hospital, Haruard Medical School, Boston, Massachusetts 02215, and Department of Medicine, Massachusetts General Hospital, Haruard Medical School, Boston, Massachusetts 021 14 Received August 3, 1992

Insulin’s mechanism of receptor binding is not well understood despite extensive study by mutagenesis and X-ray crystallography. Of particular interest are Yanomalous”analogues whose bioactivities are not readily rationalized by crystal structures. Here the structure and dynamics of one such analogue (GlyB2Cinsulin) are investigated by circular dichroism (CD) and isotope-aided 2D-NMR spectroscopy. The mutant insulin retains near-native receptor-binding affinity despite a nonconservative substitution ( P h e B 2 4 4 l y ) in the receptor-binding surface. Relative to native insulin, GlyB24-insulin exhibits reduced dimerization; the monomer (the active species) exhibits partial loss of ordered structure, as indicated by CD studies and motional narrowing of selected ‘H-NMR resonance. 2D-NMR studies demonstrate that the B-chain @-turn(residues B20-23) and @-strand(residues B24-B28) are destabilized; essentially native a-helical secondary structure (residues A3-A8, A1 3-A1 8, and B9-B 19) is otherwise maintained, 13CIsotope-edited NOESY studies demonstrate that long-range contacts observed between the B-chain @-strand and the a-helical core in native insulin are absent in the mutant. Implications for the mechanism of insulin’s interaction with its receptor are discussed. ABSTRACT:

How insulin binds to its receptor poses a problem of general interest. Do conformational changes in the hormone accompany receptor binding, and if so, what structural features are recognizedby the insulin receptor? Studiesof active analogues containing C-terminal deletions in the B-chain have focused attention on the B-chain &strand (residues B24-B28) as a possible site of conformational change (Fischer et al., 1985; Nakagawa & Tager, 1986, 1987; Casaretto et al., 1989; Schwartz et al., 1989). Evidence for the functionalimportance of flexibility in this region has been provided by studies of analogues containing cross-linksbetween the A- and B-chains (Brandenburg et al., 1972; Cutfield et al., 1981; Nakagawa & Tager, 1989; Brems et al., 1991). Of particular interest is mini-proinsulin, an inactive insulin analogue that contains a peptide bond between the B- and A-chains (LysB29-GlyAl; Markussen, 1985). Despite its complete loss of activity, the crystal structure of mini-proinsulinis identical to that of native insulin (Derewenda et al., 1991). Such discordance between function and structure suggests that the various crystal structures of native insulin (Adams et al., 1969;Peking Insulin Supportedby grantsfrom the American Diabetes Association (S.E.S. and M.A.W.), Juvenile Diabetes Foundation International (S.E.S. and M.A.W.), NSF (S.E.S.),and NIH (M.A.W.). The Harvard Medical School NMR Facility is supported by NIH 1 S10 RR04862-01. Insulin analogues were prepared with the assistanceof a Josh Diabetes Center DERC grant (DK36836). * Address correspondenceto this authorat theDepartmentof Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115. t Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School. f Permanent address: Institute of Biophysics,Academia Sinica,Beijing, China. 11 Josh Diabetes Center and Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School. * Department of Medicine, MassachusettsGeneral Hospital,Harvard Medical School.

0006-2960/92/043 1-11940$03.00/0

Structure Group, 1971; Blundell et al., 1972; Bentley et al., 1978;Smith et al., 1984;Baker et al., 1988; Derewenda et al., 1989; Badger et al., 1991) also depict inactive conformers. Derewenda et al. (1991) have thus proposed that the extent of conformationalchange in the insulin-receptor complex must exceed the range of structural variation among different crystal forms (Dodson et al., 1979; Cutfield et al., 1981; Chothia et al., 1983). To complement crystallographicstudies,we and others have used 2D-NMR’ spectroscopy to investigate the structure of insulin or insulin analogues in solution (Weiss et al., 1989; Kline & Justice, 1990; Boelens et al., 1990; Hua & Weiss, 1990,1991; Kristenson et al., 1991;Knegtel et al., 1991; Hua et al., 1991, 1992). Comparative NMR studies of analogues with enhanced or reduced affinity for the insulin receptor would be expected to provide new insights into the mechanism of receptor binding. In this paper we focus on [PheB24Gly]insulin, whose “anomalous” near-native bioactivity seems inconsistent with models of insulin receptor-binding based on crystal structures (Mirmira & Tager, 1989). Remarkably, an overall loss of ordered structure in the mutant protein is observed by circular dichroism. Correspondingperturbations in structure and dynamics are observed by T2 relaxationstudies (motional narrowing) and 2D-NMR spectroscopy. These perturbations are localized to the C-terminal region of the B-chain (residuesB20-B30) and lead in turn to global changes in the predicted receptor-binding surface. We propose that the solution structure of GlyB24-insulin provides a model for Abbreviations: CD, circular dichroism; DQF-COSY, doublequantum-filtered correlated spectroscopy;DG, distance geometry; DPI, des-pentapeptide(B26-B3O)-insulin, HMQC, heteronuclear multiplequantum coherence; NMR, nuclear magnetic resonance; NOE, nuclear Overhauser enhancement, NOESY, NOE spectroscopy; TFA, trifluoroacetyl; TOCSY, total correlation spectroscopy.

0 1992 American Chemical Society

Nonlocal Structural Perturbations in a Mutant Human Insulin

Biochemistry, Vol. 31, No. 47, 1992

11941

sites of conformational change in the hormonereceptor complex with possible application to analogue and drug design (Brange et al., 1988; Markussen et al., 1988).

MATERIALS AND METHODS Semisynthesis of GlyB24-HumanInsulin andl”C-Labeled GlyB24-HumanInsulin. t-Boc-Obenzyl-~-[ring-~~C4] tyrosine, t-Boc-~-[ring-~~Cs]phenylalanine, and t-Bot-[ 1,2-13C2]glycine were purchased from Cambridge Isotope Laboratories (Woburn, MA). Additional unlabeled protected amino acids and reagents were purchased from Applied Biosystems; solidk(nm) phase syntheses were performed on an Applied Biosystems Model 430A synthesizeras described (Shoelson et al., 1992). FIGURE1: (A) Far-UV CD spectra of native human insulin (solid Des-octapeptidehuman insulin was kindly provided by R. E. line) and GlyB24-insulin (dashed line) in aqueous-HC1 solution (pH 1.9) at 25 OC. (E)Corresponding CD spectra of monomeric insulin Chance (Eli Lilly & Co., Indianapolis, IN). Semisynthesis analogue (DKP-insulin; Weiss et al., 1991) containing PheB24 (solid of GlyB24-insulin from DO1 and the correspondingprotected line) and GlyB24 (dashed line) under similar conditions. Similar octapeptide [NH-Gly-Gly-Phe-Tyr-Pro-Lys(Boc)-Thr-OH] attenuation of mean residue ellipticity is observed a t pH 7.4. In each was accomplished in trypsin-catalyzed reaction as described case the protein concentration was 100 pM. DKP-insulin contains the substitutions HisBlOAsp, ProB28Lys, and LysB29Pro. (Inouye et al., 1979, 1982; Shoelson et al., 1983b, 1992; Nakagawa & Tager, 1986). The analogue was purified as described (Shoelson et al., 1992) and characterized by amino Solvent Conditions for NMR Study. Native insulin forms acid composition and HPLC. Incorporationof 13C-labelsinto dimers and higher-order oligomers in aqueous-HC1 solution GlyB24-insulin was verified by NMR (below). (Goldman & Carpenter, 1974). Accordingly, the present 2DCD. Spectra were obtained at 25 “C using an Aviv NMR studies are conducted primarily in 20% deuteroacetic spectropolarimeter with a 1-mm path length cuvette. The acid, conditionsunder which native insulin is monomeric and protein concentration (determined by absorbance at 280 nm) stably folded (Weiss et al., 1989; Hua & Weiss, 1991). was 25 pM; native insulin is primarily but not completely Although GlyB244nsulin was not designed to be monomeric, monomeric under these conditions. Ellipticity was measured NMR studies in &20% deuteroaceticacid (see below) indicate for 10 s at each nanometer; final spectra were smoothed by that the analogue is in fact monomeric in aqueous-HC1 solution averaging adjacent points. (pH 1A). Accordingly, ‘H-NMR studies of GlyB244nsulin NMR. Spectra were recorded at 500 mHz at Harvard may be conducted both in 20% deuteroacetic acid and in Medical School. A total of 2048 points were sampled in tz; aqueous-HC1 solution (pH 1.8). The former conditions permit 400 tl values were obtained, and the data matrix was zeroeffects of the mutation to be determined relative to native filled to 2K X 2K. TOCSY and NOESY spectra were insulin, and the latter provide a control for solvent composition. processed using shifted sine-bell window functions with exponential apodization in both dimensions; DQF-COSY ID ‘H-NMRStudies. Spectra of GlyB244nsulinandnative spectra were processed similarly, with the additionof a squared insulin in aqueous-HC1 solution (pH 1.8) and in 20% sine-bell function. 13C-Isotope-editedexperiments were perdeuteroacetic acid are shown in Figure 2 (left-hand box). We formed as described by Bax and Weiss (1987); 13C decoupling first compare the spectra of the native and mutant proteins was accomplished during lH acquisition using the WALTZin aqueous-HC1 solution. The arrow and asterisk (panels A 16 scheme. Carr-Purcell-Meiboom-Gill (CPMG) studies and D) indicate tyrosine resonances in the B-chain (B16 and, of T2 relaxation times were performed with 16 transversedelay to a lesser extent, B26) whose linewidths are sensitive to times in the range 0-1.20 s; the data was fit to a single dimerization (Weiss et al., 1989). These resonances are sharp exponential decay with the time constants as given in Tables in the spectrum of GlyB244nsulin (panel A) but broadened 111 and IV. by intermediate exchange in the spectrum of native insulin (panel D). Such differences are seen more clearly in DQFRESULTS COSY spectra; antiphase cancellationof broad resonances in (Z) Overall Features of Circular Dichroic and ‘ H - N M R the native spectrum leads to disappearanceof the TyrB16 and Spectra PheB24 aromatic cross-peaksand to attenuation of the TyrB26 cross-peak (supplementarymaterial). Such broadeningis not Circular Dichroism. In Figure 1A are shown far-UV CD observed in 20% deuteroacetic acid, in which the native and spectra of native insulin (solid line) and GlyB24insulin (dashed mutant proteins are each monomeric. line). Attenuation in mean residue ellipticity is observed at wavelengths (205-230 nm) sensitiveto formationof secondary Motional Narrowing and Chemical-Shift Dispersion. Instructure. These changes suggest substantial loss of ordered spection of the lH-NMR spectrum of GlyB244nsulin suggests structure in the mutant protein. Because partial attenuation two subclasses of spin systems: (a) reSOnanceSwhose linewidths of such CD bands can be observed on dissociation of insulin and secondary shifts are similar to those of corresponding dimers into monomers (Melberg & Johnson, 1990), control resonances in native insulin, and (b) resonances whose studies of PheB24 and GlyB24 analogues were conducted in linewidthsand secondary shifts are significantly smaller than an engineered insulin monomer (Figure 1B) (Weiss et al., those of corresponding resonances in native insulin. The 1991;Shoelson et al., 1992). Similar attenuationof structureexistence of these subclasses suggests that loss of ordered sensitive CD bands is observed, indicating that GlyB24 structure (as inferred from circular dichroism) occurs in a perturbs the structure of the monomer. GlyB244nsulin may discrete region of the protein as analyzed further below (part also exhibit reduced dimerization, as suggested by the small difference in the extent of attenuation in the two CD studies. 11).

11942 Biochemistry, Vol. 31, No. 47, I992 I

I

I

Hua et al. I

'

1

'

1

'

1

1

'

slowly -exchanging amide resonances

D

C B

'e

b CI d

9.0

A &

I

I

7.0

8.0

FIGURE 2: (Left-Hand Panel) (A) One-dimensional IH-NMR spectrum of GlyB24-insulin in aqueous solution (pH 1.8). Downfield amide resonances a 4 are assigned to A1 1, B8, B9, and B6, respectively (inset X 10). (B) IH-NMR spectrum of GlyB24-insulin in 20% acetic acid. (C) IH-NMR spectrum of native human insulin in 20% acetic acid. (D) 'H-NMR spectrum of native human insulin in aqueous solution (pH 1.8) exhibits extensive intermediate-exchangebroadening due to self-association (Weiss et al., 1989). Arrows and asterisks in panels A and D indicate resonances of TyrB26 (which overlap other aromatic resonances in each case). The B26 resonances are very sensitive to partial dimer formation,since the B26 ring packs in the dimer interface. These resonances are broadened in native insulin (D), but not in the spectrum of GlyB24-insulin (A). In each case the protein concentration was ca. 1.5 mM. (Right-Hand Panel) Slowly-exchangingamide rwonanca of GlyB24-insulin in 20% acetic acid demonstrate the stability of the a-helices over 12 h; corresponding exchange in native insulin is 3-fold slower (Hua & Weiss, 1990, 1991a,b). (A) Base-line 'H-NMR spectrum of GlyB24-insulin in 80% H20/20% deuterated acetic acid. (B) IH-NMR spectrum of GlyB24-insulin 10 min after dissolution in 80% Dz0/20% deuterated acetic acid. (C-G)Successive time points are 20 min, 40 min, 60 min, 2 h, and 12 h, respectively. Resonance assignments in panels A and B: (a) A1 1, (b) B8, (c) B9, (d) B6, (e) A12 and B19, (f) A13, (8) A2 and B18, (h) A6 and A7, (i) A4, A16, A17, B15, and B16, 0') B10, (k) B13, (1) A19 and B17, (m) A10 and B14, (n) A14 and A15, and ( 0 ) A18. Conformational Broadening of Amide Resonances. As in the 'H-NMR spectrum of native insulin (Weiss et al., 1989; Kline & Justice, 1990), differential broadening of amide resonances is observed in the spectrum of GlyB24-insulin and ascribed to exchange among conformational substates (Figure 2, left-hand panel). Presumably, collective motions in native insulin responsible for conformational broadening are retained on a similar time scale in the mutant protein. Such motions do not require the breakage of hydrogen bonds, as indicated by the observation of slowly-exchanging amide resonances (Figure 2, right-hand panel). Conformational broadening is more severe in aqueous-HC1 solution than in 20% deuteroacetic acid (Figure 2, left-hand panel) as illustrated by the four amide resonances resolved in the downfield region (CysAl 1, GlyB8, SerB9, and LeuB6; labeled a d in spectrum A).

(II) 2D ' H - N M R Studies in 20% Acetic Acid Sequentialassignment is first obtained in 20%deuteroacetic acid and extended to aqueous-HC1 solution in part 111. The primary structures of the A- and variant B-chains of GlyB24insulin are shown in Figures 3 and 4, respectively. The A-chain contains 21 residues, and the B-chain contains 30 residues. Spin-system classification is accomplished by analysis of DQFCOSY and TOCSY spectra; representative TOCSY spectra of GlyB24-insulin are shown in Figure 5 in 20% deuteroacetic acid (panels A and C) and aqueous-HC1 solution (panels B and D). In either solvent a single spin system is observed for each residue. Sequential Assignment. HN-Ha ("fingerprint") and HNHN regions of the NOESY spectrum of GlyB24-insulin in 20% deuteroacetic acid are shown in Figure 6. Sequential and medium-range contacts are shown separately for the A-chain (panels A and C) and B-chain (panels B and D). A

oquoouo oolutlon

20% ocetlc a d d Ashain d,N

dNN

(i.i+3)

---

G IV E O C C T S I C S L Y O L E N Y C N

- - -- -

-

- --

-

G I V E O C C T S IC S L Y O L E N Y C N

-

--

-

-

-

FIGURE 3: Summaryof sequential and medium-rangeconnectivities in the A-chain in 20% acetic acid (left) and aqueous solution at pH 1.8 (right). The format and symbols are as developed by Wuthrich (1982,1986). Analogous data for native insulin have been published (Hua & Weiss, 1991b). summary of sequential and medium-range NOES is provided in schematic form (Wuthrich, 1986) in Figures 3 (A-chain) and 4 (B-chain). Because fingerprint NOES may be attenuated both by amide broadening (in the folded region of the protein) and local dynamics (in unfolded regions), interpretation of their relative intensities is not straightforward. Complete assignment of 'H-NMR resonan= is obtained and given in Table I; differences from native insulin (>0.1 ppm) are given in Table 11. Motional narrowing and loss of secondary shifts involving residues B2&B30 are described in turn. ( i )Motional Narrowing. Because cross-peak intensities in DQF-COSY spectra are sensitive to the dimensionless ratio Aw/J (Weiss et al., 1984), their comparison in the spectra of native and mutant insulins provides a qualitative indication of changes in resonance linewidth and (in the absence of intermediate-exchange mechanisms) of relative mobilities. In Figure 7 are shown slices through representative DQF-COSY multiplets assigned to the B-chain. Relative narrowing is

Biochemistry, Vol. 31, No. 47, 1992 11943

Nonlocal Structural Perturbations in a Mutant Human Insulin 20% acetic acid

aqueous solution

FIGURE 4: Summary of sequential and medium-range connectivities in the B-chain in 20% acetic acid (left) and aqueous solution at pH 1.8 (right). The format and symbols are as described in Figure 3.

4.5

I

'

3.5 I

I

2.5

I

1.5 I

I

0.5 I

4.5

I

3.5 I

2.5 1

1.5

0.5 I

I

__

8 I 70

I

I

I

--

1

-

0 -

81

--a

I

I

I

I

I

I

ZO 68 FIGURE 5: Comparison of TOCSY spectra of GlyB24-insulin in 20% acetic acid (panels A and C) and in aqueous solution (panels B and D). A general correspondence of spin-system chemical shifts is observed. Selected aliphatic spin systems are labeled in panels A and B. Enlargement of the aromatic regions is shown in panels C and D with assignments indicated. 74

72

68

observed involving residues in the C-terminal region of the B-chain (B22-B30). Resonances assigned to the side chains of ValB12 and LeuB15 are also narrowed in the spectrum of the mutant protein; in native insulin these side chains pack against PheB24 and TyrB26. Their greater mobility in the mutant presumably reflects loss of such long-range side-chain interactions. In Figure 8 are shown corresponding A-chain DQF-COSY multiplets. Although cross-peak intensities are

74

7.2

similar in the two cases, relative narrowing of C-terminal a@ cross-peaks (GlnA15, AsnA18, and TyrA19) is observed in the spectrum of GlyB24-insulin. In native insulin this region packs against side chains of the B-chain a-helix (including LeuB15). Measurement of nonselective spin-spin relaxation times (T2)complements qualitative analysis of DQF-COSY multiplet intensities to provide evidence of motional narrowing.

Hua et al.

11944 Biochemistry, Vol. 31, No. 47, 1992 I

I

I

I

I

6.5 . A 7.0 8.0

7.5 8.4 8.0 8.5 9.0

0

e9.5

>I

I

A - Chain I

I

I

I

I

I

i

6.5

A

- Chain

9.6

7.2

7.0 7.5 8.0 8.0 8.4 8.5 9.0 9.5 J

3.5

I

I

I

1

I

9.0

8.5

8.0

25

20

I

6.5

4.7

4.5

4.3

4.1

3.9

3.7 3.5

FIGURE6: Sequential assignment of GlyB24-insulin in 20% acetic acid. Connectivities in the A- and B-chains are shown separately: (A) d” connectivities in the A-chain, (B) d” connectivities in the B-chain, (c)dmNconnectivities in the A-chain ‘fingerprint” region, and (D) daN connectivitiesin the B-chain ‘fingerprint” region. Additional resonances in panels A and B are labeled as follows. Panel A (a) TyrA14-Ha to -&NH, (e) TyrAl9-Ha to -%Ic, ( f ) G ~ B ~ - H , Ito N -HH I z ~(9) ~, to -He, (b) TyrB26-Ha to -He,(c) TyrB16-Ha to -He,(d) AsnA18-Hdl~~ G~~AS-H,INH to -H,zNH,(h) B3Asn-Hal~~ to -H~zNH, (i) GlnAIS-Hsl~~ to -H*zNH, (j)impurity, (k) LeuA16-NH to TyrA19-H,, and (1) TyrB16-H, to LeuB17-NH. Panel B: (m)PheBZS-Ha to -NH, (n) TyrB26-Ha to -NH, ( 0 ) TyrB16-Ha to LeuB17-NH, (p) LeuBl1-NH to HisB10-NH, (9)SerA9-NH of ThrA8-NH, (r) AsnA21-NH to CysA20-NH, (s) AsnA18-NH to GlnA17-NH, (t) not clear, (u) HisBS-Hs to -NH, (v) TyrA14-NH to LeuA13-NH, and (w) SerA12-NH to GlnAIS-NH. In such experiments it is important to compare only protons of a given type, e.g., Ala CpH3, Phe H,, etc. Carr-PurcellMeiboom-Gill (CPMG) studies of native insulin in 20% deuteroacetic acid demonstrate a large variation in transverse relaxation times (Tables I11 and IV), which for each type of proton may be classified as follows: (a) fast, protons belonging to side chains in the hydrophobic core (such as LeuB15); (b) intermediate, protons belong to surface side chains with folded backbone structure (such as HisB10); (c) slow, protons belonging to partially disordered regions of the polypeptide chain (such as PheB1). The latter resonances are assigned to theN- and C-termini (Al, A21, Bl-B3, B29-B30) and the B-chain @-turn(B2CrB23) (Boelens et al., 1990; Hua & Weiss, 1990,1991). Corresponding CPMG studies of GlyB24-insulin demonstrate that whereas T2 values of protons in the N-terminal arm (residues Bl-B9) aresimilar to those of native insulin, a systematic trend is observed among residues B22B30 toward longer relaxation times in the mutant protein (Table 111). Longer T2’s are also seen for the methyl resonances of ValB 12 and LeuBl5 in accord with the relative intensities of their DQF-COSY cross-peaks. In the CPMG series fewer resonances are resolved in the A-chain than in the B-chain (Table IV).

(ii)Pattern of Secondary Shifts. GlyB24-insulin exhibits significant differences in chemical shifts (f>O. 1 ppm) relative to native insulin (Table 11). Such differences may reflect the absence of the PheB24 ring-current field (Kristensen et al., 1991; Weiss et al., 1991) and/or structural perturbations in the mutant protein. Consistent with either mechanism, dispersion of chemical shifts in the spectrum of GlyB24-insulin is less extensive than that of native insulin under the same conditions (Hua & Weiss, 1991). The meta resonances of three tyrosines (A19, B16, and B26), for example, exhibit nearly equivalent chemical shifts in the mutant protein (Table I). Two additional observations are noteworthy. (i) Whereas the chemical shifts of residues B26B30 in native insulin differ significantlyfrom those of a random-coil peptide corresponding to residues B23-B30 (parts A and B of Table V), the chemical shifts of residues B26B30 in GlyB24-insulin are similar to those of the corresponding random-coil peptide (part C of Table V; differences in B23 and B24 chemical shifts are presumably due to the inductiveeffect of the N-terminal TFA group in the peptide). (ii) Whereas in native insulin a unique TyrB26 ortho-meta TOCSY cross-peak is observed, in the spectrum of GlyB24-insulin major (93%) and minor (7%) aromatic cross-peaks are seen (data not shown). A similar

Biochemistry, Vol. 31, No. 47, 1992

Nonlocal Structural Perturbations in a Mutant Human Insulin Table I: Chemical Shifts of Assigned 'H-NMR Resonances of GlvB24-Human Insulin in 20% Acetic Acid at 25 "C

11945

Table 11: Chemical Shift Differences of GlyB24-Insulin and Native Insulin in 20% Acetic Acid at 25 OC (>0.1 ppm) ~

~~

chemical shifts at 25 "C residue A1 Gly A2 Ile A3 Val A4Glu A5 Gln

NH

C"H

C@H

others

3.97, 3.80 1.13 8.47 3.86 8.03 3.62 8.08 4.20 8.20 4.05

O H 2 1.13, 0.93 O H 3 0.74, C6H30.63 C7H3 0.90,0.85 1.92 2.06,2.06 C7H2 2.48,2 48 2.06: 2.01b CYH2 2.46,2.38 N'H2 6.83, 7.50 2.86: 3.31b 3.73: 3.306 C7H3 1.22 4.39 4.01.3.86 C7H2 1.09,0.44 1.53 C7H3 0.63, C*H30.51 3.31,2.86 3.97: 4.28b 1.34, 1.37 CYH 1.41, C6H30.78,0.71 2.96: 2.88b C2,6H 7.02, C3,5H, 6.79 C7H2 2.42, N'H2 6.94, 7.47 2.33,2.00 1.88, 1.62 CYH 1.70, C6H30.82,0.78 C7H2 2.56, 2.29 2.09,2.00 2.57,2.49 N*Hz 6.50, 7.13 2.97: 3.29b C2,6H 7.29, C3,5H 6.73 3.25,2.84 2.85,2.77 3.14, 3.14 C2,6H 7.19, C3,5H 7.33 C4H 7.25 O H 3 0.82,0.82 1.88 2.71, 2.71 N6H2 7.53,6.91 2.07: 1.88b CYH2 2.23, 2.15 N'H2 7.34,6.79 3.23," 3.52b C2H 8.57, C4H 7.40 1.73: 0.93b C'H 1.56, C*H30.90,0.74 3.22,2.93

A6 Cys A7 Cys A8Thr A9 Ser A10 Ile

8.26 8.27 8.19 7.39 7.78

4.88 4.82 4.05 4.74 4.37

A l l Cys A12Ser A13 Leu A14Tyr A15Gln A16Leu A17Glu A18Asn A10Tyr A20 Cys A21 Asn B1 Phe

9.60 8.71 8.60 7.48 7.55 8.06 8.06 7.43 7.89 7.44 8.17

4.88 4.57 3.85 4.14 3.95 4.14 4.13 4.45 4.48 4.68 4.74 4.27

B2Val B3 Asn B4Gln

8.13 4.12 8.44 4.69 8.38 4.46

B5 His B6 Leu B 7 Cys B8 Gly B9 Ser B10 His B11 Leu B12 Val B13 Glu B14 Ala B15 Leu B16 Tyr B17 Leu B18 Val B19 Cys B20 Gly B21 Glu B22 Arg

8.58 8.93 8.29 9.20 9.02 8.00 7.10 7.21 7.94 7.79 8.13 8.10 7.88 8.46 8.67 7.75 7.99 7.97

B23 Gly B24 Gly B25 Phe

8.11 3.97, 3.92 8.03 3.88, 3.88 2.97,2.89 7.91 4.59

4.44 4.48 4.94 3.98, 3.84 4.10 4.51 4.03 3.42 4.07 4.09 3.98 4.17 4.06 3.87 4.72 3.88, 3.88 4.28 4.30

B26 Tyr 8.03 4.58 B27 Thr 7.83 4.50 B28 Pro 4.36 B20 Lys 8.25 4.37

3.87, 3.87 3.56: 3.2gb 1.86: 1.24b 2.01 2.14,2.05 CPH3 1.48 1.71, 1.41 3.10, 3.10 1.64,b 1.89b 2.14 2.85: 3.18b 2.12,2.01 1.96, 1.82

2.89,2.89 4.03 2.26, 1.90 1.83, 1.75

C2H 8.69, C4H 7.46 CYH 1.37, C6H30.83,0.76 C7H3 0.93,0.85 C'H2 2.51 CYH 1.63, C6H30.75,0.68 C2,6H 7.05, C3,5H 6.72 CTH 1.81, C6H30.91,0.91 C'H3 1.03,0.81 C7H2 2.48, 2.41 CYHz 1.66, 1.66 C6H23.20, 3.20 N'H 7.18 C2,6H, 7.07, C3,5H 7.22 C4H 7.16 C2,6H 6.99, C3,5H 6.73 C'H3 1.16 C"H2 1.97, 1.97, C6Hz3.63 C'H2 1.47, C6H2 1.65 C'H2 2.96, N'H 7.48 CYH3 1.15

4.38 B30 Thr 8.04 4.46 Chemical shifts are measured relative to the residual CH3 resonance of acetic acid, presumed to be 2.03 ppm. bj3 resonances for which stereosmific assignment has been obtained.

ratio of major and minor cross-peaks is observed in spectra of the synthetic B23-B30 octapeptides. We ascribe such states to trans and cis isomers of the ThrB27-ProB28 peptide bond. In native insulin the trans configuration is stabilized by packing interactions with the A-chain (Baker et al., 1988; Hua & Weiss, 1991). Secondary-Structure Analysis. Observation of nested (i,i+3) connectivities and associated strong d" NOES demonstrates the existence of three a-helices in the solution

chemical shifts at 25 OC residue B11 Leu B12 Val B14Ala B15 Leu B17Leu B18 Val B21 Glu B22 Arg B23 Gly B24 Gly B25Phe B26 Tyr B28 Pro B30 Thr

NH

CaH

CSH

others CYH0.11

0.14 0.11 0.13 0.10

0.16

0.13

O H 0.22, C6H30.13,0.25

0.10 -0.37 0.47

0.10 0.13 0.13 -0.31

-0.10

-0.29 0.11 -0.27 0.13

structureof GlyBZCinsulin: A3-A8, A1 3-A1 8, and B9-Bl9. Similar helix-relatedNOEs (A2-A8, A12-Al8, and B9-Bl9) are observed in DPI and native insulin (Hua & Weiss, 1990, 1991). The significance of the small difference in A-chain helical end points is unclear. In native insulin residues B20B23 adopt a (1+4) @-turn,which permits the C-terminal @-strand(B24-B26) to pack against the B-chain a-helix (B9B19). The presence of the @-turnin native insulin is indicated by local NOESreflecting the geometry of the @-turn(contacts are observed between ArgB22-HN and the a protons of GlyB20, and between ArgB22-Hg1,2 and the a proton of CysB19) and nonlocal NOES reflecting the B-chain a/@ orientation (contacts are observed from PheB24 to ValB12 and TyrB16, and from TyrB26 to LeuB11, ValB12, and LeuBl5). None of these NOES are observed in the spectrum of GlyB2Cinsulin, suggesting that the @-turnis not stably maintained. Long-Range NOEs and Tertiary Structure Analysis. The tertiary structure of native insulin consists of an a-helical core and adjoiningC-terminal &strand of the B-chain (residues B24-B28). The folding of these elements in GlyB24-insulin is described in turn. ( i ) Tertiary NOEs Define a Native a-Helical Core. The relative orientations of the three a-helices in GlyBZCinsulin are defined by long-range NOEs, which are analogousto those observed in the spectrum of native insulin (Hua & Weiss, 1991). Representative NOES involving aromatic rings and aliphatic resonances are shown in Figure 9 (the boxed region in panel A is provided in expanded form in the supplementary material). Contacts areobserved between IleA2 and TyrAl9, between ValA3 and TyrA19, and between IleAlO and HisB5. A2-Al9 NOES demonstrate the proximity of the N- and C-terminal a-helices of the A-chain; A10-B5 NOEs define a local packet between the A-chain and N-terminal (nonhelical) region of the B-chain. Also observed in this region of the NOESY spectrum are (i,i+3) helix-related NOES between LeuA16 and TyrA19 and (i,i+4) NOES between ValB12 and TyrB16. Interchain contacts are also observed in each case between the side chains of TyrA19 and LeuB15. Interchain contacts between AsnA2l and ArgB22 Hgpredicted by crystal structures of native insulin and observed in solution as an AsnA2 1-N,/ArgB22 Hb NOE (Hua & Weiss, 1991)-are not observed in the NOESY spectrum of GlyB24insulin. The absence of this NOE indicates either a stable increase in the A21-B22 distance (Le., >4.3 A) or the introduction of subnanosecond fluctuations of the local A21B22 interproton vectors of sufficient amplitude to quench the NOEs (Wuthrich, 1986).

++ ++ ++ +-+ Hua et al.

11946 Biochemistry, Vol. 31, No. 47. 1992 Mulanl

Native

Mutant

T y r 816(08I

L Y SBZ9ly61

Native

Mutant

T y r A19 (981

Ser AS(a81

Gln AIS(aB1

Val A3(eBl

Native

Val B I Z la81

m: 10

T h r 8 2 7 1001

I le A2 ( y z 8 I

-",--+

"14

Tyr 826 (a01

+a-

Leu B6(yIlI

+%-Leu 86(s81

Arg B Z Z I s P l

++ Gln B4(a81

FIGURE7: Analysis of B-chain DQF-COSY cross-peak intensities. One-dimensional slices are shown through resolved multiplets in the 2D spectra of GlyB244nsulin (left-hand column of each panel) and native human insulin (right-hand column of each panel). Assignments are as indicated. In the mutant protein motional narrowing is evident in the B20-B30 region and those side chains contacted by this region innativeinsulin (ValB12andLeuBl5;seeFigure9).Thetwospectra were normalized according to the a@ cross-peak of TyrA14 (a flexible residueon thesurfaceofthe A-chain remote fromthesiteof mutation). For clarity corresponding cross-peaks were scaled by 1/ m as indicated (Le., intense signals have low m value, and weak signals have high m value).

(ii) 13C-Isotope-Edited 2D-NMR Studies Demonstrate Disruption of Contacts between the B-chain @-Strand and Hydrophobic Core. In native insulin the aromatic rings of

FIGURE8: Analysis of A-chain DQF-COSY cross-peak intensities. The format is the same as in Figure 7. Some A-chain motional narrowing in GlyB24-insulin is seen, presumably as a transmitted effect of the B-chain perturbation. No narrowing is seen for IleAlO or CysA7, which form an A-chain/B-chain contact remote from the site of mutation. Table 111: Selected SpinSpin Relaxation Times Tz' of B-Chain Resonances residue B1 B3 B5 B10 B12 B15 B18 B22 B22 B24 B25 B26 B26 B28 B29 B30 B27/B30"

proton He HR He Hs Cy2H3 C6ZH3 C6lH3 HR Hs

Hs

Ha Hu Ha Ha He Hu C7H3

native insulin 360 60 390 440